5G Synchronization Aspects Michael Mayer Senior Staff Engineer Huawei Canada Research Centre WSTS, San Jose, June 2016 Page 1
Objective and outline Objective: To provide an overview and summarize the direction being taken with 5G mobile networks, focussing on what is required in terms of synchronization (what, when, how, and why) Outline: 5G: what is it? Review of modulation Some architectural components What may be required to time this Page 2
5G: WHAT IS IT Page 3
4G as the starting point UE (phone) enodeb S 1 enodeb MME S 1 S 1 X 1 X 1 X 1 Core network enodeb LTE-A is the starting point Also defined as IMT-Advanced 5G is also known as IMT-2020 Mobility is key, but typical advanced service is the Smart Phone Evolution of previous wireless networks to fully packet based Different types of cells (macro base station, pico-cell, etc.) User data rates in the order of 10Mbit/s to support traditional services (voice, web, video) Synchronization is typically supplied from the core over the S1 interface Page 4
LTE features Core network LTE-A technology also includes MIMO MME Carrier aggregation UE (phone) enodeb S 1 enodeb S 1 S 1 X 1 X 1 X 1 enodeb HetNets Home Base station All increase performance to the end user. End user is typically using a smart phone. How can this architecture be further developed to address evolving services? Page 5
5G drivers: need anything else be said? Type of characteristics of network devices are also changing, phone and tablet Still dominate (61% as of Q1/15), but growth in M2M(24%) and Car (15%) Source: MOBILE DATA DEMAND: GROWTH FORECASTS MET Significant Growth Projections Continue to Drive the Need for More Spectrum, CTIA, Page 6
5G networks 5G networks will: Support of new services: IoT, Sensor Networks, tactile internet With higher performance: Higher bit rates, higher speed mobility handoff, lower latency Networks deployed with: Higher connection densities, new spectrum allocation, use of unlicensed spectrum There is a need to look at the synchronization requirements Page 7
5G/IMT-2020 Timeline You are here New elements to offer capabilities of IMT-2020 Vision Requirements Standards development Standards enhancement Systems deployment * Other radio systems IMT-200and IMT-advanced and their enhancement Spectrum for IMT Evolution/Integration with other radio systems Systems deployment Enhancement and related development of standards (Rec. ITU-R M.1457 and ITU-R M.2012) Spectrum implementation 2000 ~ 2014 2015 2016 2017 2018 2019 2020 The sloped dotted lines in systems deployment indicate that the exact starting point cannot yet be fixed. ~ : Possible spectrum identification at WRC-15 and WRC-19 * : Systems to satisfy the technical performance requirements of IMT-2020 could be developed befor e year 2020 in some countries. : Possible deployment around the year 2020 in some countries (including trial systems) Sync standards development Network backhaul: ITU-T SG-15 Radio: 3GPP M.2083-05 Page 8
Capability summary Enhanced mobile broadband Gigabytes in a second 3D video, UHD screens Smart home/building Work and play in the cloud Augmented reality Smart city Voice Future IMT Industry automation Mission critical application Self driving car Massive machine type communications Ultra-reliable and low latency communications M.2083-02 Source: ITU-R Rec.M.2083; Usage scenarios of IMT for 2020 and beyond Page 9
Capability delta Area traffic capacity 2 (Mbit/s/m ) 10 100 Network energy efficiency 1 10 Peak data rate (Gbit/s) 0.1 1 20 10 6 10 5 Connection density 2 (devices/km ) Figure: ITU-R M.2083 1 IMT-2020 IMT-advanced User experienced data rate (Mbit/s) 100 10 10 1 Latency (ms) 1 350 400 3 500 Spectrum efficiency Mobility (km/h) M.2083-03 5G realized by: Peak data rate New spectrum allocation User data rate CoMP, MIMO Spectrum efficiency Modulation Mobility Faster hand-off capability (CRAN) Latency Movement of service data bases Connection density CRAN architecture Network energy CRAN architecture Area traffic capacity CRAN/CoMP/MIMO Page 10
5G use case families and related examples (NGMN) Page 11
Synchronization requirements? Synchronization requirements will be based on two aspects: the needs of the service and the needs of the infrastructure Service: M2M, IoT may require accurate synchronization, which may or may not be provided by the network. Infrastructure needs New air interfaces may be defined New capabilities related to time-sensitive networks may be needed and may require synchronization support from the network Page 12
OFDM transmission and reception Key to understanding the air interface sync requirement Page 13
Review of modulation Modulation gets the bits on/off the air Wireless modulation is generally one form of Orthgonal Frequency Division Multiplexing (OFDM) Splits (multiplexes) high data rate bit streams onto multiple sub-carriers, each with narrow bandwidth Used in: LTE-A: WIMAX Attraction for wireless: Low symbol rate (1/15kHz) is less susceptible to ISI Subcarrier equalization simpler due to narrow subcarrier BW Use with QAM results in high spectral efficiency Well understood (!) Page 14
Recall OFDM f 0 OFDM Transmitter f 1 f 2 f 3 f 4 f 5 f 6 f n High bit rate in (single carrier) Multiple low bit rate out (multiple carriers form the symbol ) f 0 f 1 f n Low bit rate OFDM improves performance for multi-path environments, key to wireless Page 15
Basic LTE frame (FDD) 10 ms duration, 10 sub-frames 1 2 3 4 5 6 7 8 9 10 Resource element: -Contains one OFDM symbol -May represent data or control -Consists of symbol plus Cyclical Prefix (CP) (i.e. 15 ksamples/s) Slot=0.5 ms Resource block: Contains seven OFDM symbols for 12 sub-carriers Page 16
TDD Frame Page 17
Visualizing for multiple Antennas Sub-Carrier In the case of MIMO, each antenna will have a frame structure. Where multiple antennas share the same system clock, usually not an issue. CoMP or Carrier Aggregation may need some assistance where antennas are driven by different oscillators. UE enodeb Page 18
OFDM in practice Data S/P Inverse FFT P/S D/A, CP* Add OFDM symbol stream Tx Channel Data P/S FFT CP Remove S/P A/D Rx *CP: Cyclical Prefix added to reduce ICI Timing recovery/generation Page 19
What goes wrong? The process of timing extraction at the receiver involves multiple steps. Need to preserve timing relationship of individual subcarriers when symbols are carried over a complex channel Process generally involves coarse frequency acquisition, followed by fine timing adjustment Carrier Frequency Offset: error in recovering the carrier, which leads to loss of orthogonality of individual sub-carriers, and increased BER But, Doppler is also a contributor. The user can be moving in a high-speed train. (5G is targeting 500 km/hr) Page 20
Extension to MIMO Timing offset (TO) and carrier frequency offset (CFO) are compensated by the receiver In a single antenna case, only one pair of oscillators, so compensation relatively straightforward. LTE-A frame format appears to be effective in supporting pilotaided synchronization and channel estimation. MIMO: Potential for multiple oscillators and therefore multiple offsets that need to be corrected. Excess time error leads to the possibility of interference with multiple carriers Field reports indicate that current methods are effective, but suggest that tight external synchronization be provided Page 21
NEW ARCHITECTURE CONSTRUCTS: CRAN AND NETWORK SLICING Page 22
New architectural components Centralization of functions and the use of data centre techniques is seen as key to supporting 5G requirements Architecture: C-RAN (cloud-ran, Centralized-RAN) Localizes the functions to allow sharing of resources Network slicing Controlling how the end-to-end network is shared between different user (groups) to achieve different network objectives Involves aspects of virtualization and orchestration Page 23
Recall the evolving base-station model Air interface (frequency, jitter) S 1 Data + timing Back haul enodeb Control UE (phone) Starting point Air interface S 1 Data + timing REC Front haul RE UE (phone) Separate antenna From enodeb (e.g. CPRI link) In practice, a network may have hundreds or thousands of base stations. Are there any efficiencies that can be gained by centralizing? Page 24
C-RAN Control Boundary clock (T-BC) Network interface Radio interface Antenna Network interface Packet Network (e.g. Ethernet) (two layer, Lossy) Radio Interface interface Server Server Server Antenna C-RAN: Localize functions associated with radio base Station control with the goal of achieving statistical gains. Data centre model Allows tighter control of latency and synchronization in the case of distributed MIMO, or CA, where normally separate base-stations would have been deployed Above example shows timing distribution via boundary clocks, but other possibilities exist. Synchronization impact: consistent with existing HRM for time/phase/frequency Page 25
Network slicing 5G networks will aggregate multiple services on possibly separate radio access technologies Spectrum may be shared and needs control and coordination Network slice represents the portion of all network resources that may be allocated to a service or user. Synchronization impact: Slicing orchestration needs to consider synchronization of air Interface. Page 26
Slicing: Allocate/connect resources Air interface Air Slice Orchestration Interconnect Storage Compute Compute Resources Connectivity/Transport Antenna resources Storage Resources Legend: RAN 1 RAN 2 RAN 3 Page 27
Summary 5G will offer new services and support new technology. Some services may require accurate time (e.g. location accuracy) New architectures for 5G appear to be well supported by current synchronization techniques. Some further work may be required to define certain components and potentially new air interface definitions. Page 28
Thank you www.huawei.com Page 29